We make our tools; our tools make us

The title to this could have come from Marshall McCluhan, a man fascinated by tools that he calls "media," and their influence on civilization. 

One example McCluhan cites: When Gutenberg began printing Bibles, this meant the Roman Catholic hierarchy no longer could monopolize access to the sacred text of the Bible. Vernacular transations followed and were printed, and that ensured there would (later) be a reformation (and a counter-reformation) re-making the structure of authority in Europe. This ultimately led to the downfall of feudalism, the enlightenment, and influenced the founding ideas leading to the USA. The same phenomenon also led to the counter-reformation, the Inquisition, and many wars. So...a mixed blessing.

More recently, Yuval Noah Harari says AI has hacked the "glue" of civilization (trust) Computers can now undermine the authority of bureaucracies of politicians, lawyers and bankers. Dangerous!

 

Ultimately, the destinations of machine learning depend on existing learning, and there are plenty of reasons to believe that existing learning is imperfect. One TED talk describes the ways we make bad decisions.  

There is even evidence that the algorithms that govern our physical existence, making sure we can digest food, walk upright, etc. are subject to "bugs" - flaws that have managed to avoid the elimination assured by evolution. See "Super-Normal Stimuli" for examples that include both humans and animals. 

One example: refined sugar evades the messages our digestion sends with other kinds of food to let us know we've consumed enough. You can suck on a "Big Gulp" sugary drink all day, and you'll never feel full. The epidemic of obesity we currently experience is one of the products of a bug in a digestive algorithm. 

Other bugs include the seven deadly sins (pride, wrath, envy, greed, gluttony, sloth and lust) and the apparently infinite appetite for safety, if not justice. Calling "sin" a software bug describes why it's to be avoided. The wages of sin are death. Bugs often crash the computers running the bad code.

So AI may also exploit humanity's buggy software even though "Why AI Doesn’t Think, Cannot Reason, Isn’t Intelligent and Will Never Achieve Consciousness" is a thing. “When God wants to drive a person insane, he grants that person’s every wish.” (Paulo Coelho). 

Meanwhile: Dave Dayen reports "It's becoming the case that it costs more for a business to deploy AI and fire human workers than it does to keep the human workers around. That could be an economic death knell for the whole technology. "

Still some pretty useful items from AI. For example: AI Detects Hidden Warning Signs Before Major Earthquakes (SciTech Daily) 

An introduction to the positive side of machine learning (AI) describes how, after mastering games, from simple to Chess and Go, AI now does medical research, correlating things like protein production to DNA/RNA structures, and creating proteins that fit with the ones discovered like this. The lecturer says such discoveries would take PhD students literally years, while computers can do the calculation in hours, if not minutes. The speaker is humble, and willing to admit computers need guardrails, but the benefits appear real.

 

AI is at least a mixed blessing. 

Meanwhile, the attack on the poor continues (The Poor Get Prison: The Alarming Spread of the Criminalization of Poverty). Current economic priorities discount the idea that NI (natural intelligence) could beat AI, and suppress the diversity of opinions and points of view. Sure, Newton spent hours on alchemy, and even more on theology, so humans can, like AI, "hallucinate." But we got modern physics from Newton (and, Planck, Bohr, Shroedinger, Einstein, etc.).